Constraint Imperative Programming

点击翻译按钮获取中文摘要

This project continues work on Constraint Imperative Programming (CIP) languages. A constraint represents a relation that should be maintained by a computer program. Constraints have a number of desirable properties that make them useful in building systems such as interactive graphical applications. They are declarative statements, rather than imperative statement; a programmer need only to state ``what'' rather than ``how''. Additional advantages result from embedding support for constraints in the programming language itself. First, it becomes convenient for programmers to use constraints. Second, the language implementation can ensure that the constraints are checked automatically. Third, a more uniform and elegant language can result from expressing standard constructs as constraints. Finally, more opportunities for optimization are opened to the compiler. In this project, the activities include investigating techniques for compiling object-oriented constraint imperative languages, such as Kaleidoscope, to produce efficient code. To accomplish this, the system does as much compile time analysis of the constraints as possible. Other activities include developing tools for writing, browsing, and debugging constraint imperative programs, investigating the compilation of Kaleidoscope programs to run on shared memory MIMD machines, and testing the utility of this class of languages by programming a set of representative applications. Another important focus is the development of a more efficient and powerful constraint satisfaction algorithm. A final activity is work on debugging support for constraint systems. The work on constraint satisfaction algorithms and debugging is useful both to support constraint imperative languages, as well as other kinds of constraint-based systems.